TechLinked Podcast Summary

Episode: Macbook Neo, M5 Pro/Max, OpenAI backtracks, Meta Glasses Video leaks + more!
Date: March 5, 2026
Host: Linus Media Group

Episode Overview

This episode dives into a packed week in tech news, highlighting Apple’s surprising hardware launch blitz (including the MacBook Neo and M5 Pro/Max upgrades), OpenAI’s controversial Pentagon contract and subsequent PR scramble, Meta’s privacy-flaunting Ray-Ban AI glasses, and several rapid-fire updates on Windows 12 rumors, Android 16 desktop mode, AI-driven RAM shortages, major iPhone exploits, and robots on Xiaomi’s EV assembly line.

Key Discussion Points & Insights

1. Apple’s Hardware Avalanche

[00:28–03:45]

MacBook Neo Announcement:
- Powered by an A18 Pro chip (almost identical to the iPhone 16 Pro, minus one GPU core, but still packs 6 CPU cores)
- Starts with 8 GB of unified memory:
  - “Sadly, the paltry eight gigabytes of unified memory.” (B, 00:38)
- 256 GB storage for $599—an unusually affordable MacBook entry point
- Host quips that Apple’s silicon prowess has long been ‘overkill’ for most users:
  - “I will be a bit shocked if this entry level MacBook isn’t enough for most people.” (B, 01:17)
- Tongue-in-cheek “Matrix” and “AppleSauce” puns tie into product naming and Tim Cook’s persona
Pro/Max MacBook Pros with M5:
- 14- and 16-inch Pro models get M5 Pro & M5 Max upgrades—now with more cores, memory, storage, and a “Fusion Architecture” (chiplet design)
- New and confusing CPU core naming:
  - “Efficiency cores are the same, what used to be called performance cores are now called super cores, and in between them there’s a new type...taking the performance core designation from what are now called super cores. Do you understand?” (B, 02:22)
  - TL;DR: They’re very powerful.
MacBook Air, Displays, and Apple’s Event Pace:
- New M5 MacBook Air—standout mention: starts with 16 GB RAM and 512 GB storage, a pointed “callout” to previous criticism.
- New Studio display and Studio Display XDR briefly acknowledged.
- Host jests that Apple launched six products without an actual themed event:
  - “The Cook was feeling saucy this week. Get it? That's a two for one pun. Tim Cook. Applesauce.” (B, 03:29)

2. OpenAI’s Pentagon Contract & Backlash

[03:45–05:32]

Backstory:
- OpenAI CEO Sam Altman renegotiates Pentagon contract after major user/employee backlash.
- Anthropic refused Pentagon contract over safety guardrails; OpenAI quickly announced acceptance—seen as opportunistic.
Fallout:
- Massive spike in ChatGPT uninstalls (+295%), with competing Claude app shooting to #1 on the App Store.
Altman’s Public Rebuttal:
- Internal memo (also posted on Twitter): new contract language to “explicitly prohibit use of OpenAI’s models for domestic surveillance of Americans.”
- Altman claims:
  - “The Department of War has agreed to OpenAI’s red lines on mass surveillance and autonomous weapons.” (B, 04:34)
- But sources (The Verge, former policy researchers) allege the Pentagon “never actually budged.” The contract merely defaults to current U.S. law, which broadly justifies surveillance.
- Notable quote on Altman’s checkered track record:
  - “So to sum up, the guy whose first company got acquired after he lied about how many users there were, then got fired...then started the nonprofit that stole the entire Internet...then arguably stole Scarlett Johansson’s voice for ChatGPT, then argued that AI training is fine because it uses less energy than human life, is continuing to be less than forthright with the public, no?” (B, 05:05)
Overall Tone:
- Highly skeptical and irreverent toward OpenAI’s corporate PR and Altman’s credibility.

3. Meta Ray-Ban AI Glasses Leak Private Videos

[05:32–06:41]

Investigative Findings:
- Swedish newspapers reveal Meta has sent “intimate user footage” from Ray-Ban AI Glasses to human moderators in Kenya.
- Reviewed videos include private activities (e.g., on the toilet, undressing, intimacy).
Moderator Testimony:
- “One worker told reporters that they regularly see everything from banking details to naked bodies. All the things you’d want.” (B, 06:02)
Privacy Policy Concerns:
- Reporters had to dig for Meta’s wearable privacy policy; main advice: don’t record anything sensitive.
- Meta’s AI terms: company reserves right to review all interactions, manually or automatically.
- Host’s summary:
  - “Don’t record it if you don’t want a stranger to see it.”
- With new facial recognition features coming, the host laments an increasingly dystopian, surveillance-heavy future.
- Direct call-out:
  - “What’s that? You’d prefer AI enabled products to have rigorous security standards and to explicitly promise not to train AI models on your data? Well, psh then, that’s our sponsor.” (B, 06:35)

4. Quick Bits (Rapid Tech Headlines)

[07:34–11:05]

Windows 12 Rumors Debunked

Flood of “Windows 12 releasing this year” headlines traced back to a mistranslated and ultimately debunked German article.
“Windows Central stepping in to quickly debunk the mess, confirming Windows 12 is absolutely probably not happening this year. Great news, as Microsoft still needs at least a few years to figure out how to squeeze ads into the volume slider.” (B, 08:00)

Android 16’s Desktop Mode Officially Rolls Out

Now available to Pixel phones (post-Pixel 2); enables full desktop experience via USB-C DisplayPort.
“Now you can proudly tell your boss you’re working from your phone…even though you could get more work done because your phone is a desktop now. Oh man, that'll get ‘em so good.” (B, 08:36)

AI Drives DDR5 RAM Scalping

Bots fire millions of requests to scrape/buy DDR5 RAM, leaving smaller companies fighting “scalp and nail.”
“Can’t wait for Michael Bay’s Aliens versus Scalper Predators, rated R for restricted memory.” (B, 09:18)

Koruna iPhone Exploit Unveiled

A severe exploit kit with 23 vulnerabilities, targeting iOS 13–17, likely used by state actors (U.S., Russia) and criminal campaigns.
“It’s unlikely these hackers will target your parents, who’d rather eat lead than update their phones, but you might wanna try convincing them one more time, just in case.” (B, 10:16)

Xiaomi Puts Robots on Its EV Assembly Line

Robots reach 90% success over hours of human-paced work, but viewed as “interns” compared to human labor.
“Here in the west, we’re worried robots will be faster…workers in China get so much done that the robots have to intern first just to keep up.” (B, 11:05)

Notable Quotes & Memorable Moments

On Apple’s entry-level MacBook pricing:
- “At a time when computer prices are threatening to get on a big rocket and leave the solar system…” (B, 00:59)
On confusing Apple core nomenclature:
- “Efficiency cores are the same, what used to be called performance cores are now called super cores, and in between them there’s a new type…” (B, 02:22)
On OpenAI’s credibility:
- “The guy whose first company got acquired after he lied about how many users there were…is continuing to be less than forthright with the public, no?” (B, 05:05)
On Meta’s data use policies:
- “Don’t record it if you don’t want a stranger to see it.” (B, 06:32)
On robot productivity in China vs the West:
- “Workers in China get so much done that the robots have to intern first just to keep up.” (B, 11:05)

Timestamps for Key Segments

MacBook Neo and Apple Hardware Announcements: 00:28–03:45
OpenAI and Pentagon Contract Controversy: 03:45–05:32
Meta Ray-Ban Glasses Privacy Breach: 05:32–06:41
Quick Bits (Windows 12, Android 16 desktop, RAM scalping, Koruna exploit, Xiaomi robots): 07:34–11:05

Tone

Witty, irreverent, and skeptical—hosts riff on industry news with running puns, sarcasm, and a relentless focus on the absurdities and contradictions present in tech industry behavior.

This summary covers all critical points, highlights the podcast’s memorable humor, and offers a clear guide for those who want to catch up on March 5, 2026’s top tech headlines—no listening required.

Transcript

A (0:01)

This episode is brought to you by Indeed. Stop waiting around for the perfect candidate. Instead, use Indeed Sponsored Jobs to find the right people with the right skills fast. It's a simple way to make sure your listing is the first candidate. C According to Indeed data, Sponsored Jobs have four times more applicants than non sponsored Jobs. So go build your dream team today with Indeed. Get a $75 sponsored job credit@ Indeed.com podcast. Terms and conditions apply.

B (0:28)

Have you ever wanted tech news like so bad that just the idea of waiting while some guy he just kind of was talking, he just goes and there's unrelated thing. So how did and how did that feel Apple's week of announcements continued this morning with the MacBook Neo powered by an A18 Pro, the same chip used in the iPhone 16 Pro models. What? Or almost the same. The Neo loses one GPU core, but keeps the six CPU cores and sadly, the paltry eight gigabytes of unified memory. But hey, Agent Smith also underestimated Neo the guy, not the MacBook. And look what happened to him. Melted into goo. I saw the movies. Really? The key detail Here is the MacBook Neo starts with 256 gigs of storage for $599 at a time when computer prices are threatening to get on a big rocket and leave the solar system. The power of Apple's chips has essentially been overkill for their devices for a while, so I will be a bit shocked if this entry level MacBook isn't enough for most people. Speaking of overkill though, Apple also announced that 14 and 16 inch MacBook Pros can now be configured with the brand new M5 Pro and M5 Max, which not only feature more cores, memory and storage, but but also a new chiplet based design we're supposed to call Fusion Architecture. If we don't want to incur Tim Cook's wrath gnashing of teeth. That was Tim Cook, by the way. Apple has also done some confusing things with their CPU cores. Efficiency cores are the same, but what used to be called performance cores are now called super cores. And in between them there's a new type of core that is taking the performance core designation from what are now called super cores. Do you understand? Look TL. Dr. The chips are powerful enough to make more of whatever is going on with this beautiful mesmerizing CGI goo. And that's good enough for me. There's also a new M5 MacBook Air whose promo video specifically mentions starting with 16 gigabytes of unified memory and 512 gigabytes of storage in what seems like a very topical callout. Current events. Then there's also a new studio display and Studio display xdr, which seem cool, but we don't even have time to talk about them. Six products during a non Apple event. The Cook was feeling saucy this week. Get it? That's a two for one pun. Tim Cook Applesauce Connor okay, OpenAI CEO and Boots sommelier Sam Altman is claiming he's renegotiating OpenAI's Pentagon contract in response to backlash from users and employees. The negative response was prompted by what happened last Friday, when, hours after the government blacklisted Anthropic for refusing to let the Pentagon use Claude without safety guardrails, OpenAI swooped in and announced its own deal with the Department of War. Naturally, this went over great, with ChatGPT uninstalls surging 295% in the fallout and Anthropic's Claude shooting to number one on the app Store. Altman eventually admitted the original deal seemed opportunistic and sloppy in an internal memo that he also posted to Twitter stating now that new contract language would explicitly prohibit the use of OpenAI's models for domestic surveillance of Americans. Scam. I'm sorry, Sam Altman is now claiming that the Department of War has agreed to OpenAI's red lines on mass surveillance and autonomous weapons. But according to multiple sources who spoke to the Verge and many other people, the Pentagon never actually budged on the contract terms. OpenAI's contract simply defers to existing US laws that have historically been used to justify mass surveillance. OpenAI's own former head of policy research said employees should assume that OpenAI caved and then framed it as not caving. So to sum up, the guy whose first company got acquired after he lied about how many users there were, then got fired from being president of Y Combinator for conflicts of interest, then started the nonprofit that stole the entire Internet, then withheld information from his board about launching ChatGPT, then arguably stole Scarlett Johansson's voice for ChatGPT, then argued that AI training is fine because it uses less energy than human life, is continuing to be less than forthright with the public, no? An investigation by Swedish newspapers SVD and Gutbergsposten sure has found that Meta has been sending intimate user footage from its Ray Ban AI glasses to human moderators in Kenya. Kenya, of all places. Workers at Meta subcontractor Sama said they've reviewed videos of people wearing these glasses on on the toilet, undressing and being intimate with each other, which is impressive. It can't be easy attracting a mate while wearing one of these things. One worker told reporters that they regularly see everything from banking details to naked bodies. All the things you'd want. Engadget noted that reporters had to jump through a bunch of hoops just to find Meta's wearable privacy policy, which puts the onus on users to not share sensitive information. Meta's AI terms of service explicitly state that Meta retains the right to review your interactions with its AIs, and that review may be automated or manual. The only solution Meta offers is to tell users not to share information on these recordings. I guess that you don't want the AIs to use and retain, which basically translates to don't record it if you don't want a stranger to see it. And with Meta's upcoming name tag feature adding facial recognition to the glasses, this mass data collection doesn't bode well for human beings who don't like being mass surveilled in a Neal Stephenson esque dystopia. What's that? You'd prefer AI enabled products to have rigorous security standards and to explicitly promise not to train AI models on your data? Well, psh then, that's our sponsor.